Serial blockface electron microscopy (SBFEM) is revolutionizing the mapping of neural microcircuits. Small volumes of brain can be fully reconstructed at nanometer-scale resolution, providing a complete description of the form and location of all synaptic inputs to a single cell. For very local inputs, the identity (and synaptic inputs and outputs) of these presynaptic cells can also be reconstructed. This technical breakthrough parallels the transformative impact of genetically modified animals and viruses for characterizing and manipulating molecularly defined neuronal cell types. Using viruses, cre-lox technology, optogenetics, and chemogenetics, distributed functional circuits can be imaged, mapped, activated, silenced or deleted. This proposal seeks to bridge the divide between these approaches. We propose to perfect a method we have devised exploiting molecular-genetic technologies to mark defined cell types with an electron-dense label for SBFEM analysis. Specifically, we have generated a recombinant adeno-associated virus (AAV) that delivers a Cre-dependent genetic construct to infected cells. Exclusively in Cre-expressing cells, the viral payload expresses a membrane-targeted marker protein comprising a fusion of a fluorescent protein (membrane-targeted green fluorescent protein - mGFP) to a recombinant peroxidase enzyme (APEX2). Pilot data show that infected Cre-expressing cells strongly express the fusion protein throughout the membrane (soma, dendrites, axons and terminals). Its bright fluorescence permits detailed confocal analysis; enzyme histochemical processing reveals the same structures by electron-dense marking visible after SBFEM sectioning and imaging. Already, the method has great promise for targeted `connectomic' analysis of Cre-expressing neurons in any brain region and of their output synapses in remote neural structures. Here, we aim to improve and extend the method. Our aims for this proposal are: 1) to optimize the histochemical protocols and design of viral constructs to mark specific cellular structures or compartments without masking synaptic vesicles and other organelles; and 2) to expand the potential applications of the method, by restricting the fluorescent and ultrastructural labeling t neurons that innervate specific targets; and by generating a knock-in mouse line that expresses the marker through mating to a Cre driver line or injection of Cre-expressing viruses.